0206

STRONG-MOTION PARAMETERS: DEFINITION, USEFULNESS AND

PREDICTABILITY
Julian J BOMMER1 And Alejandro MARTNEZ-PEREIRA2

SUMMARY
Estimating strong-motion characteristics for seismic risk assessment and earthquake-resistant
design requires the clear definition of parameters that reflect the destructive potential of the
motion. There is currently considerable ambiguity in the definition of many parameters, in
particular duration, and unqualified statements are frequently repeated regarding the usefulness of
certain parameters. Current approaches to predicting the values of these parameters do not
facilitate the rational estimates of the most severe earthquake motions.
INTRODUCTION
The principal objective of engineering seismology is to provide quantitative estimates of expected levels of
seismic ground-motion as the basic input to earthquake-resistant design, the evaluation of collateral seismic
hazards, such as liquefaction and landslides, and seismic risk assessment. This invariably entails characterising
the complex nature of strong-motion accelerograms through the use of simple parameters and the development
of predictive relationships for these parameters. Since the first strong-motion accelerograms were obtained in the
Long Beach earthquake of 1933, a large number of parameters have been defined to characterise salient features
of the ground motion. In recent years, increasing numbers of parameters have been proposed for this purpose,
many of them of increasing complexity. However, careful study of the technical literature in this field reveals
that there is often considerable ambiguity or disagreement regarding the definition of even the simplest strongmotion parameters. This paper addresses this issue, discussing the definition of peak acceleration parameters and
duration and the consequences of the variations, often not specified that exist amongst the definitions employed.
The usefulness of strong-motion parameters is dependent primarily upon their intended use. The parameters that
can be employed in earthquake-resistant design are few and are directly related to the methods of structural
analysis used in current practice. For the purposes of seismic risk assessment, through empirical loss functions
relating structural damage levels to a given measure of ground-motion intensity, in theory any parameter can be
employed. The usefulness of the parameter in this case will depend purely on the degree to which it is a measure
of the destructive potential of the motion. It is shown that very few parameters, if any, can effectively
characterise the nature of strong motion in isolation and many, by themselves, convey nothing about the
destructive potential of the shaking. It is also shown that although there is often poor correlation between simple
strong-motion parameters and damage levels, it is often possible to identify lower bound values for damage to
occur.
Once a strong-motion parameter has been selected to characterise the ground motion, it is necessary to develop
relationships between values of this parameter and features of the earthquake source, the travel path to the site
and nature of the site. The nature of such predictive relationships for duration depends very heavily on the
definition of duration employed. Even for a new and robust definition of duration, it is found that reliable
prediction is difficult because of the dependence on features of the seismic source that cannot be predicted for
future earthquakes. The way in which strong-motion parameters related to amplitude are predicted is also
examined and important questions are raised regarding current approaches, particularly with respect to measures
of mean values and associated scatter. The possibility of employing estimates of upper bounds on the values of
some parameters is discussed.
1
2

Imperial College of Science, Technology & Medicine

Imperial College of Science, Technology & Medicine

DEFINITIONS OF STRONG-MOTION PARAMETERS

The most widely used parameter in strong-motion studies is the peak ground acceleration (PGA). Since
accelerograms are triaxial records of the ground motion, the definition of PGA depends on how the two
horizontal components are handled. Campbell (1985) lists various approaches to handling the horizontal
components of accelerograms in defining PGA, the most common being the larger of the two values, the
randomly oriented component (determined using both components as independent data points) and the mean of
the two peaks. The second two approaches should generally yield almost identical mean values from the
predictive relationships but it should be noted that the use of the mean artificially lowers the standard deviation
of the regression. Boore et al. (1993) derive attenuation relationships, using the same data set, for both the larger
and randomly oriented components, the former predicting values as much as 20% higher than the latter. The
same options also exist for any other amplitude-based parameter of the horizontal motion, including response
spectral ordinates for which all of the above options have been used as well as performing regressions on the
spectral ordinates determined from the vectorially resolved maximum component (Kawashima et al., 1984).
Since the total duration of an accelerogram depends on either the diligence of the digitiser, for analogue records,
or the pre- and post-event intervals, for digital records, it is not possible to define the duration of strong shaking
as simply the time between the start and finish of an accelerogram. Many definitions of strong-motion duration
have been presented in the literature, all of which attempt to isolate a certain portion of the accelerogram during
which the strongest motion occurs. It has been found that all of these definitions can be classified into one of
four generic categories (Bommer & Martnez-Pereira, 1996). The first category is bracketed duration, being
simply the interval between the first and last excursion of a particular threshold amplitude. The second category
is uniform duration, which is the sum of all of the time intervals during which the amplitude of the record is
above the threshold. The third category is significant duration, which is determined from the Husid plot (Husid,
1969), usually based on the interval during which a certain portion of the total Arias intensity is accumulated.
The fourth generic category is structural response duration, determined by applying one of these three
categories of definition to the response of a specific single-degree-of-freedom oscillator. The thresholds used to
define the duration, of whichever generic category, can be either absolute values of acceleration or Arias
intensity, or else relative values defined as a proportion of the maximum value. A final distinction can be made
between definitions based on the entire accelerogram and those determined after passing the record through a
narrow-band filter to isolate a particular frequency. Table 1 lists a total of 38 studies that have proposed
definitions for strong-motion duration and classifies them according to the categories that have been described.
For a given accelerogram, the durations determined according to different definitions can vary by more than a
factor of two. Furthermore, for weak accelerograms the definitions based on absolute criteria can result in zero
duration, whereas the definitions based on relative criteria, being related only to the geometry of the
accelerogram, will always yield a value of duration (Bommer & Martnez-Pereira, 1999).
USEFULNESS OF STRONG-MOTION PARAMETERS
The strong-motion parameter most widely used in seismic design is the acceleration response spectra, although
there is an increasing tendency towards the use of displacement and energy spectra (Fajfar, 1999). This move
away from force-based seismic design is driven by the recognition that structural damage during earthquakes is
not controlled by accelerations and hence damage can be limited more effectively by controlling either
displacements or energy dissipation. For spectrum-based design methods, the type of input is limited by the
method of analysis although these new trends ensure a better reflection of the destructive capacity of the motion.
In time-history analysis, it is important that the accelerograms employed have appropriate values of those
parameters that are closely correlated with damage capacity.
It is widely recognised that it is very difficult to characterise the nature of strong ground-motion with a single
parameter, but at the same time there are other concepts regarding damaging features of strong motion which are
widely repeated and yet unfounded. For example, it is often stated that the destructive capacity of ground motion
increases with its duration, although by itself the duration gives no indication at all of the damage potential. For
two accelerograms with similar values of PGA, the one with longer duration is likely to be more destructive, if
the frequency contents are similar. However, strong-motion recordings from earthquakes in El Salvador in 1982
(Ms 7.3) and 1986 (Ms 5.4) revealed that in both cases the Arias intensity of the motion in San Salvador was
almost identical. However, the smaller, local event produced shaking with a duration equal to about one tenth of
the duration of the shaking from the larger, more distant earthquake. The transmission of the same amount of

0206

energy in a much shorter interval of time resulted in considerably greater damage in the 1986 earthquake
(Bommer et al., 1997).

Table 1 Classification of definitions of strong-motion duration

Bracketed
Uniform
Definition of Duration
Rel
Abs
Rel
Abs
Rosenblueth & Bustamente (1962)
z
Housner (1965)
z
Ambraseys & Sarma (1967)
Lynch (1967)
Husid (1969)
z
Kobayashi (1971)
z
Page et al. (1972)
Donovan (1972)

Bolt (1973)
Housner (1975)
Trifunac & Brady (1975)
z
OBrien et al. (1978)
z
Hisada & Ando (1976)
z
Apitkaev (1977)
Trifunac & Westermo (1977)
Saragoni (1977)
z
z
McGuire & Barnhard (1979)
McCann & Shah (1979)
Perez (1980)
Takizawa & Jennings (1980)
Vanmarcke & Lai (1980)
Zahrah & Hall (1984)
Zhou & Xie (1984)
Shahabi & Mostaghel (1984)
Theofanopoulos & Drakapoulos (1986)
Elghadamsi et al. (1988)
Xie & Zhang (1988)
Mohraz & Peng (1989)
Theofanopulos & Watabe (1989)
z
z
Kawashima & Aizawa (1989)
Ellis et al. (1990)
z
Sarma & Casey (1990)
z
Papazachos et al. (1992)
Bommer & Martnez-Pereira (1996)
Somerville et al. (1997)
z
Wembo & Kezhong (1997)
Safak (1998)
Bommer & Martnez-Pereira (1999)
z frequency-independent duration
frequency-dependent duration

Significant
Rel
Abs

Structural
Rel
Abs

z
z

z
z
z

z
z
z

z
z

z
z
z
z

z
z

z
z

It is also often stated that PGA is a very poor indicator of the damage capacity, which is true, and that there is
very poor correlation between observed damage and recorded values of PGA. However, using a carefully
compiled dataset of 391 strong-motion recordings and 30 isoseismal maps, pairs of values of MMI and PGA
were extracted and plotted in Fig. 1 (Martnez-Pereira, 1999). Notwithstanding the shortcomings of using
intensity as a measure of damage, it can be clearly seen from the plot that for the motion to be potentially
damaging to engineered structures (MMI VIII) the value of PGA must be at least 0.2g. In this way, lower
bounds for potentially damaging parameters have been established for many parameters, including 20 cm/s for
peak ground velocity (PGV) and 0.8m/s for Arias intensity (AI).

0206

10

PGA (g)

0.1

0.01
4

10

11

12

MMI

Figure 1. Peak ground acceleration (PGA) versus Modified Mercalli intensity (MMI).
An equally important observation that can be made from Fig. 1 is that there are clearly many accelerograms with
PGA greater than 0.2g but which are clearly not damaging, being associated with intensity of shaking as low as
VI or even V on the Modified Mercalli scale. If other thresholds, based on other characteristics of the ground
motion, were also applied, for example that the PGV be greater than 20 cm/s and the damage index defined by
Fajfar et al. (1990) be greater than 30, then many of the weak motions with PGA greater than 0.2g would be
removed. The ability of any single parameter to measure the damage potential of the ground motion will depend
to a significant degree on which features of the motion (amplitude, frequency content, duration and energy) it
reflects; it is likely that any parameter which reflects any three of these properties will implicitly reflect the
fourth as well. However, no damage potential parameter is of any use unless it is emprically or analytically
correlated with indices or measure of damage, or at least a threshold value for damage is specified.
PREDICTABILITY OF STRONG-MOTION PARAMETERS
Seismic hazard assessment involves estimation of the ground motions to be produced by future earthquakes. This
is normally achieved through the use of attenuation relationships that predict values of a selected strong-motion
parameter as a function of earthquake parameters such as magnitude and distance from the seismic source to the
site. There are very fundamental difficulties encountered with current approaches to predicting strong-motion
parameters and these are discussed here with reference to the prediction of duration and of amplitude-based
parameters such as PGA and response spectral ordinates.
Prediction of Strong-Motion Duration
The nature of attenuation relationships for strong-motion duration is very much dependent on the definition of
duration employed. All the relationships based on durations defined by relative levels on the accelerogram or
Husid plot predict values of duration that increase indefinitely with distance, whereas those based on absolute
levels of acceleration predict durations that decrease with distance. The effective duration, DE, defined by the
authors of this paper (Bommer & Martnez-Pereira, 1999), is found to correlate well with the moment
magnitude, Mw. Using 32 accelerograms from rock sites located at distances of less than 10 km from the
earthquake source, in order to decouple the influence of distance and soil layers from the dependence on
magnitude, the following relationship was found:

log( D E ) = 0.69 M w 3.70

(1)

The standard deviation of this regression is 0.28. Using the empirical relationship between rupture length and
moment magnitude of Wells & Coppersmith (1994) and assuming a constant fault rupture velocity of 2.5 km/s, a
relationship for the duration of fault rupture, DR, is obtained which is identical except that the constant becomes
3.62 instead of 3.70. However, the fit in Eq.(1) is only obtained by first applying correction factors for the
durations of those earthquakes caused by non-unilateral fault rupture; for earthquakes with bilateral rupture, such
as Loma Prieta (1989) and Kobe (1995), the duration of shaking would be expected to be half as long and hence
a factor of 2 was applied to the durations calculated for accelerograms from these earthquakes. This means that
although strong-motion duration can be estimated with some confidence for a given magnitude, there will always
be an uncertainty of a factor of 2 since it is not possible to know whether future earthquakes will be caused by
bilateral or unilateral rupture.

0206

Estimates of Strong-Motion Amplitudes

Other than situations in which there are exceptional soil effects, such as in Mexico City in 1985, or topographical
effects, such as ridge response at the site or Moho bounce due to path effects, earthquake motion that can cause
damage to well-engineered structures is limited to within a few tens of kilometres of the source. The magnitudedistance (M-R) space in which damaging motions can be expected has been identified using a large dataset of
accelerograms and defining as potentially damaging motions those passing the minimum thresholds discussed
previously for PGA, PGV, AI, spectral intensity and the damage index of Fajfar et al. (1990), although the M-R
space is almost unchanged using any 3 of these parameters (Martnez-Pereira, 1999). This limits damaging
motion to earthquakes of magnitude 5 or greater and at distances which reach a maximum of 30 km at M 7.7 for
rock sites and 100 km at M 7.9 for soil sites. Earthquake motions recorded at distances beyond this near-field
boundary are unlikely to be of engineering significance, but generally no more than 50% of the accelerograms in
the datasets used to derive attenuation relationships lie within these limits.
The values of strong-motion parameters are usually estimated from attenuation relationships that predict mean
values of the parameters as a function of M, R, site conditions and sometimes other characteristics of the
earthquake such as fault rupture mechanism. A typical form for such an attenuation function is:

log( Y ) = C1 + C 2 M + C 3 R + C 4 .log( R )

(2)

The problem arises in the way the regressions are performed using the logarithmic form of the equation as in
Eq.(2), which results in records of small amplitude having the same influence on the coefficients as the few
genuinely strong motions. As a result, the relationships are constrained primarily by weak records and yet the
most important application is in determining the strong earthquake motion that can be generated within a few
tens of kilometres of the source. The fact that these predictive equations have become known as attenuation
relations, with the emphasis therefore on the decay of amplitude with distance, suggests that the focus has not
been governed by engineering concerns. The recognition that strong-motion amplitudes close to the earthquake
source are not reliably predicted by standard attenuation relationships has led to derivation of correction factors
that take account of the effects of rupture directivity (Somerville et al., 1997). The application of these factors
certainly improves near-field estimates of ground-motion amplitudes, but the factors rely on the a priori
determination of source parameters which, like the non-unilateral rupture that alters duration, are difficult to
estimate for future events.
Another problem associated with the use of the logarithmic form of the attenuation model is the measure of the
scatter represented by the standard deviation &, which is measured on the logarithm of amplitude. A typical value
of & for PGA is 0.26; the mean-plus-one-standard deviation value for a small PGA of 0.05g is 0.09g, the scatter
not altering the fact that the motion is too weak to be of significance in the design of new structures. However,
the mean-plus-one-standard-deviation value of a near-field PGA of 0.50g is 0.91g, and the design implications of
the difference are vast. In probabilistic seismic hazard assessment, the scatter is truncated at some limit, but the
current options for selecting the truncation level have no physical basis. It is possible that this results in small
and distant earthquakes exerting undue influence on the hazard at a site and it definitely does not provide rational
estimates of near-field amplitudes. Since it is generally not possible to predict, for future earthquakes, the source
characteristics that give rise to the most extreme amplitudes, the assessment of extreme seismic loading may
benefit from assessment of upper bounds on amplitude parameters. It is very interesting to note that until the
early 1970s many studies were published (e.g. Brune, 1970; Ambraseys, 1974) proposing, on the basis of
physical models, limiting values on strong-motion parameters, several of which seem to have been confirmed by
subsequent recordings (Figure 2). It is equally interesting to note that in 1971 the San Fernando earthquake
produced three times as many accelerograms than had been accumulated in California over the previous 38
years: researchers could not resist the temptation of fitting curves through the clouds of data points that had
become available, despite the very small amplitudes of most of the records. From then on the focus switched
almost completely from the estimation of upper bounds to the estimation of mean values associated with huge
scatter, which has not been significantly decreased by the expansion of strong-motion recording. As the number
of accelerograms grows with expanding strong-motion networks, especially with automatic digitisation facilities
and digital accelerographs, although strong records are being obtained, the proportion of weak motions in the
databank is actually growing. The mean values of PGA predicted for a distance of 1 km from an M 7 earthquake
by Californian attenuation relationships published over the last 30 years have gradually decreased, whereas the
scatter has remained almost constant (Martnez-Pereira, 1999).

0206

The development of performance-based design, which aims to provide greater control over structural and nonstructural damage, presents new challenges for engineering seismology that may require new approaches. The
parameters that control damage must be identified and unambiguously defined. When estimating ground motions
for the most severe loading cases for which standard occupancy structures must be designed against collapse,
upper bounds may provide a more rational approach than extrapolation across large scatter. Whichever approach
is adopted, when the focus is on the most severe loading, there is strong argument for discarding a large
proportion of the strong-motion dataset, which is too weak to be relevant.

0 .5
0 .0
1 9 30

1 9 40

1 9 50

1 9 60

1 9 70

1 9 80

Petrolia

Nahanni

Gazli

Pacoima

Morgan
Hill

El Centro

1 .5

Long Beach

PGA (g)

Hokkaido
Honshu
Parkfield
Koyna

Ho rizo n t al

2 .0

1 .0

Imperial
Valley

Vert ical

Northridge

2 .5

1 9 90

Parkfield

80
60
40

El Centro

100

Long Beach

PGV (cm/sec)

120

Petrolia
Northridge

Pacoima

140

Mxico

Ye ar

20
0
1930

1940

1950

1960

1970

1980

1990

Ye ar

Figure 2. Maximum recorded values of PGA and PGV (Martnez-Pereira, 1999).

REFERENCES
Ambraseys, N.N. (1974) Dynamics and response of foundation materials in epicentral regions of strong
earthquakes. Proceedings 5WCEE, Rome, 1, cxxvi-cxlviii.
Ambraseys, N.N. & S.K. Sarma (1967) Response of earth dams to strong earthquakes. Gotechn. 17, 181-283.
Apitkaev, F. (1977) The correlation between MM-intensity, relative duration of shaking and peak acceleration.
EERI Newsletter 3, Earthquake Engineering Research Institute.
Bolt, B.A. (1973) Duration of strong ground motions. Proceedings 5WCEE, Rome, 1, 1304-1313.
Bommer, J.J. & A. Martnez-Pereira (1996) The prediction of strong-motion duration for engineering design.
Proceedings 11WCEE, Acapulco, Paper No. 84.
Bommer, J.J. & A. Martnez-Pereira (1999) The effective duration of earthquake strong motion. Journal of
Earthquake Engineering 3, 127-172.
Bommer, J.J., A. Udas, J.M. Cepeda, J.C. Hasbun, W.M. Salazar, A. Surez, N.N. Ambraseys, E. Buforn, J.
Cortina, R. Madariaga, P. Mndez, J. Mezcua & D. Papastamatiou (1997) A new digital accelerograph nework
for El Salvador. Seismological Research Letters 68, 426-437.
Boore, D.M., W.B. Joyner & T.E. Fumal (1993) Estimation of response spectra and peak accelerations from
western North American earthquakes: an interim report. US Geological Survey Open-File Report 93-509.

0206

Brune, J.N. (1980) Tectonic stress and the spectra of seismic shear waves from earthquakes. Journal of
Geophysical Research 75, 4997-5009.
Campbell, K.W. (1985) Strong motion attenuation relations: a ten-year perspective. Earthq. Spec. 1, 759-804.
Donovan, N.C. (1972) Earthquake hazard for buildings. Building Science Series 46, 82-111.
Elghadamsi, F.E., B. Mohraz , C.T. Lee & P. Moayyad (1988) Time-dependent power spectral density of
earthquake ground motion. Soil Dynamics & Earthquake Engineering 7, 15-21.
Ellis, G.W., M. Srinivasan & A.S. Cakmak (1990) A program to generate site dependent time histories: EQGEN.
Technical Report NCEER-90-0009, National Center for Earthquake Engineering Research.
Fajfar, P. (1999) Trends in seismic design and performance evaluation approaches. Proceedings Eleventh
European Conference on Earthquake Enginering: Invited Lectures, Bisch et al. (eds), Balkema, 237-249.
Fajfar, P., T. Vidic & M. Fischinger (1990) A measure of earthquake motion capacity to damage medium-period
structures. Soil Dynamics & Earthquake Engineering 9, 236-242.
Hisada, T. & H. Ando (1976) Relation between duration of earthquake ground motion and the magnitude.
Kajima ICT, Japan.
Housner, G.W. (1965) Intensity of ground shaking near the causative fault. Proceedings 3WCEE, 1, 81-94.
Housner, G.W. (1975) Measures of the severity of ground shaking. Proceedings US National Conference on
Earthquake Engineering, Ann Arbor, 25-33.
Husid, L.R. (1969) Caractersticas de terremotos anlisis general. Revista del IDIEM 8, 21-42, Santiago, Chile.
Kawashima, K. & K. Aizawa (1989) Bracketed and normalised durations of earthquake ground acceleration.
Earthquake Engineering & Structural Dynamics 18, 1041-1051.
Kawashima, K., K. Aizawa & T. Takahashi (1984) Attenuation of peak ground motion and absolute acceleration
response spectra. Proceedings 8WCEE, San Francisco, 2, 257-264.
Kobayashi, Y. (1971) Effects of earthquakes on ground. Journal of Physics of the Earth 19, 231-241.
Lynch, R.D. (1967) A statistical relationship between band-pass filter and pseudo-relative velocity spectra.
Atomic Energy Commission Report NVO-1163-188.
Martnez-Pereira, A. (1999) The characterisation of near-field earthquake ground-motions for engineering
design. PhD Thesis, Imperial College, University of London.
McCann, M.W. & H.C. Shah (1979) Determining strong-motion duration of earthquakes. Bulletin of the
Seismological Society of America 69, 1253-1265.
McGuire, R.K. & T.P. Barnhard (1979) The usefulness of ground motion duration in prediction of severity of
seismic shaking. Proceedings Second US National Conference on Earthquake Engineering, Stanford, 713-722.
Mohraz, B. & M.M. Peng (1989) The use of low-pass filter in determining the duration of strong ground motion.
Journal of Pressure Vessels and Piping Division ASME 182, 197-200.
OBrien. J.L., J.R. Murphy & J.A. Lahoud (1977) The correlation of peak ground acceleration amplitude with
seismic intensity and other physical parameters. Report NUREG-0143, US Nuclear Regulatory Commission.
Page, R.A., D.M. Boore, W.B. Joyner & H.W. Coulter (1972) Ground motion values for use in seismic design of
trans-Alaska pipeline system. US Geological Survey Circular 672.

0206

Papazachos, B.C., Ch.A. Papaioannou, V.N. Margaris & N.P. Theodulidis (1992) Seismic hazard assessment in
Greece based on strong motion duration. Proceedings 10WCEE, Madrid, 2, 425-430.
Perez, V. (1980) Spectra of amplitudes sustained for a given number of cycles: an interpretation of response
duration for strong-motion earthquake records. Bulletin of the Seismological Society of America 70, 1943-1954.
Rosenblueth, E. & J.I. Bustamante (1962) Distribution of structural response to earthquakes. Journal of the
Engineering Mechanics Divisions ASCE 88, EM3, 75-106.
Safak, E. (1998) 3D response spectra: a method to include duration in response spectra. Procc 11ECEEE, Paris.
Saragoni, G.R. (1977) The  method for the characterization of earthquake accelerograms. Proceedings
6WCEE, New Dehli, 1, 357-362.
Sarma, S.K. & B.J. Casey (1990) Duration of strong motion earthquakes. Proceedings Ninth European
Conference on Earthquake Engineering, Moscow, 10-A, 174-183.
Shahabi, M. & N. Mostaghel (1984) Strong ground motion duration and effective cyclic acceleration.
Proceedings 8WCEE, San Francisco, 2, 843-850.
Somerville, P.G., N.F. Smith, R.W. Graves & N.A. Abrahamson (1997) Modification of empirical strong ground
motion attenuation relations to include the amplitude and duration effects of rupture directivity. Seismological
Research Letters 68, 199-222.
Takizawa, H. & P.C. Jennings (1980) Collapse of a model for ductile reinforced concrete frames under extreme
earthquake motions. Earthquake Engineering & Structural Dynamics 8, 117-144.
Theofanopoulos, N. & S. Drakopoulos (1986) A study of strong ground-motion duration of earthquakes in
Greece. Proceedings Eighth European Conference on Earthquake Engineering, Lisbon, 1, 3.2/65-67.
Theofanopulos, N.A. & M. Watabe (1989) A new definition of strong motion duration and comparison with
other definitions. Structural Eng./Earthquake Eng. 6, 111s-122s.
Trifunac, M.D. & A.G. Brady (1975) A study on the duration of strong earthquake ground motion. Bulletin of
the Seismological Society of America 65, 139-162.
Trifunac, M.D. & B.D. Westermo (1977) A note on the correlation of frequency dependent duration of strong
earthquake ground motion with the Modified Mercalli intensity and the geological conditions at the recording
site. Bulletin of the Seismological Society of America 67, 917-927.
Vanmarcke, E.H. & S-S.P. Lai (1980) Strong-motion duration and RMS amplitude of earthquake records.
Bulletin of the Seismological Society of America 70, 1293-1307.
Wembo, Z. & P. Kezhong (1997) Estimation of the duration of strong ground motion. Mitigating the impact of
impending earthquakes, Vogels & Brandes eds., Balkema, 149-156.
Xie, L. & X. Zhang (1988) Engineering duration of strong-motion and its effects on seismic damage.
Proceedings 9WCEE, Tokyo-Kyoto, 2, 307-312.
Zahrah, T. & W.J. Hall (1984) Earthquake energy absorption in SDOF structures. Journal of Structural
Engineering Division ASCE, 110, ST8, 1757-1772.
Zhou, Y. & L. Xie (1984) A new definition of duration of strong ground motion. Proceedings Chinese National
Conference on Earthquake Engineering, Shangai (in Chinese).

0206